Full-field Ultrasonic Data Analysis Based on Statistical Covariance Method

Abstract

Laser ultrasonic techniques (LUTs) are becoming widely used for damage detection based on full-field ultrasonic data. In general, LUTs provide threedimensional (3-D) ultrasonic signals from which significant informative features may be extracted about the health condition of a structure. In this paper, statistical convariance method is proposed to investigate 3-D ultrasonic features. A laser ultrasonic interrogation method based on a Q-switched laser scanning system was used to interrogate 3-D ultrasonic signals in to a 2 mm aluminum plate. The ultrasonic signals in three-dimensional (3-D) space were generated, indexed by spatial N (xdirection), spatial M (y-direction), and temporal K (t-direction) samples, respectively. Then, two sets of covariance matrices were obtained based on the vector variables in x-direction and y-direction respectively for all k time samples respectively. The covariance imaging and the corresponding variance map were obtained. The wavelength and group velocity of S0 and A0 modes were able estimated from the variance map. The wavelength of the S0 and A0 modes were estimated with the deviation error of 2.3% and 11% respectively. The group velocities of S0 and A0 modes were estimated with the deviation error of 1.8% and 0.9%. The analysis will be further performed to develop group velocity curves for a range of frequency in metal and composite plates.